Tilburg University
Presurgical identification of patients with glioblastoma at risk for cognitive impairment
at 3-month follow-up
Rijnen, Sophie; Butterbrod, Elke; Rutten, Geert-Jan; Sitskoorn, Margriet; Gehring, Karin
Published in: Neurosurgery DOI: DOI:10.1093/neuros/nyaa190 Publication date: 2020 Document VersionPublisher's PDF, also known as Version of record Link to publication in Tilburg University Research Portal
Citation for published version (APA):
Rijnen, S., Butterbrod, E., Rutten, G-J., Sitskoorn, M., & Gehring, K. (2020). Presurgical identification of patients with glioblastoma at risk for cognitive impairment at 3-month follow-up. Neurosurgery, 87(6), 1119-1129.
https://doi.org/DOI:10.1093/neuros/nyaa190
General rights
Copyright and moral rights for the publications made accessible in the public portal are retained by the authors and/or other copyright owners and it is a condition of accessing publications that users recognise and abide by the legal requirements associated with these rights. • Users may download and print one copy of any publication from the public portal for the purpose of private study or research. • You may not further distribute the material or use it for any profit-making activity or commercial gain
• You may freely distribute the URL identifying the publication in the public portal Take down policy
Presurgical Identification of Patients With
Glioblastoma at Risk for Cognitive Impairment at
3-Month Follow-up
Sophie J. M. Rijnen, PhD∗ ‡ Elke Butterbrod, MSc∗ Geert-Jan M. Rutten, PhD, MD‡ Margriet M. Sitskoorn, PhD∗ Karin Gehring, PhD∗ ‡∗Department of Cognitive Neuropsych-ology, Tilburg University, Tilburg, Noord-Brabant, The Netherlands;‡Department of Neurosurgery, Elisabeth-TweeSteden hospital, Tilburg, Noord-Brabant, The Netherlands Correspondence: Sophie J. M. Rijnen, PhD, Department of Cognitive Neuropsychology, Tilburg University, Room 202, P O Box 90153, Warandelaan 2, 5000 LE Tilburg,
Noord-Brabant, The Netherlands. Email:s.j.m.rijnen@uvt.nl Received, August 13, 2019. Accepted, March 18, 2020.
C
Congress of Neurological Surgeons 2020.
This is an Open Access article distributed under the terms of the Creative Commons Attribution-NonCommercial-NoDerivs licence (http://creativecommons.org/licenses/ by-nc-nd/4.0/), which permits non-commercial reproduction and distribution of the work, in any medium, provided the original work is not altered or transformed in any way, and that the work is properly cited. For commercial re-use, please contact
journals.permissions@oup.com
BACKGROUND:Pre- and postoperative cognitive deficits have repeatedly been
demon-strated in patients with glioblastoma (GBM).
OBJECTIVE:To identify presurgical risk factors that facilitate the identification of GBM
patients at risk for postoperative cognitive impairment.
METHODS:Patients underwent neuropsychological assessment using Central Nervous
System Vital Signs 1 d before (T0) and 3 mo after surgery (T3). Patients’ standardized scores on 7 cognitive domains were compared to a normative sample using one-sample
z tests. Reliable change indices with correction for practice effects were calculated to
assess cognitive changes in individual patients over time. Logistic regression models were performed to assess presurgical sociodemographic, clinical, psychological, and cognitive risk factors for postoperative cognitive impairments.
RESULTS:At T0, 208 patients were assessed, and 136 patients were retested at T3. Patients
showed significantly lower performance both prior to and 3 mo after surgery on all cognitive domains compared to healthy controls. Improvements and declines over time occurred respectively in 11% to 32% and 6% to 26% of the GBM patients over the domains. The regression models showed that low preoperative cognitive performance posits a significant risk factor for postoperative cognitive impairment on all domains, and female sex was a risk factor for postoperative impairments in Visual Memory.
CONCLUSION:We demonstrated preoperative cognitive risk factors that enable the
identi-fication of GBM patients who are at risk for cognitive impairment 3 mo after surgery. This information can help to inform patients and clinicians at an early stage, and emphasizes the importance of recognizing, assessing, and actively dealing with cognitive functioning in the clinical management of GBM patients.
KEY WORDS:Cognition, Glioblastoma, Individual performance, Neuropsychological assessment, Risk factors
Neurosurgery 0:1–11, 2020 DOI:10.1093/neuros/nyaa190 www.neurosurgery-online.com
G
lioblastoma (GBM) accounts for the majority of gliomas (56.6%), and comprises the most aggressive primary brain tumors in adults.1-4 Although the neurooncological field is evolving, GBM is still incurable and associated with overall poor outcomes: the median overall survival despite aggressive surgical resection, radiotherapy, and chemotherapy remains∼15 mo, anddeteriora-ABBREVIATIONS: ASA,American Society of Anesthesiologist;BH,Benjamini-Hochberg;CNS VS,Central Nervous System Vital Sign;ES,effect size;GBM,glioblastoma;HADS,Hospital Anxiety and Depression Scale;KPS,Karnofsky Performance Status;NPA,neuropsychological assessment;OR,odds ratio;QoL,quality of life;RCI,reliable change index;ROC,receiver operating characteristic;SD,standard deviation;SE,standard error
tions in cognition and quality of life (QoL) over time are common.5-7
Although cognitive functioning is proven to be an independent predictor of survival in GBM patients,8,9 it is still rarely considered (ie, assessed, monitored, or treated in rehabil-itation programs) in neurooncological care. Furthermore, the literature on cognitive functioning in glioma patients is characterized by
fairly small-scale studies and strong heterogeneity in patient samples.9,10 Previous studies show extensive preoperative, and new and worsened postoperative cognitive deficits in more than 80% of GBM patients.6,10-17Sociodemographic (ie, age, sex, and education),18,19 clinical (ie, hemispheric tumor location, frontal involvement, physical health status, and tumor volume),17-20 psychological (ie, anxiety and depression),21and/or cognitive (ie, preoperative neuropsychological functioning)22factors may play a role in predicting the outcome at 3-mo follow-up in glioma patients. However, risk factors that facilitate the identification of patients at risk for postoperative cognitive impairments already before surgery have remained unknown to date. We assessed cognitive functioning in a large sample of GBM patients using a computerized neuropsychological battery 1 d before and 3 mo after surgery. The aim of the current study was to identify patients who are at risk for postoperative cognitive impairment using preoperative factors, based on sociodemographic, clinical, psycho-logical, and cognitive characteristics.
METHODS
DesignThe present study comprised a prospective longitudinal design in which brain tumor patients admitted for surgical resection at the Elisabeth-TweeSteden hospital (Tilburg, the Netherlands) between November 2010 and November 2018 underwent neuropsychological assessment (NPA) 1 d before surgery (T0) and 3 mo after surgery (T3) as part of standard neurooncological care.
All patients provided written informed consent. Study approval was issued by the Medical Ethics Committee (file number NL41351.008.12). The patient sample of the current study includes patients who were also included in a previous study.6
Patients
The patients included in the current study were those who underwent initial surgical resection, and who were diagnosed with a histopatho-logically confirmed GBM. Patients were excluded if they (1) were younger than 18 yr, (2) had a history of intracranial neurosurgery, (3) had other major medical illnesses in the past year prior to surgery (eg, myocardial infarct), (4) lacked a basic proficiency in Dutch, (5) were unable to undergo the NPA due to severe visual, motor, or cognitive problems, and/or (6) when surgical complications occurred (eg, intracranial hematoma).
Measures and Procedure
Sociodemographic Characteristics
Patients underwent NPAs per protocol, also including a checklist and standardized interview at baseline for obtaining and verifying sociodemo-graphic information (eg, age, educational level). The Dutch Verhage scale was used to classify the completed level of education (from unfinished elementary school to a university degree): Verhage 1 to 4 represent a low educational level (primary level education or lower), Verhage 5 a middle educational level (completion of average level secondary education), and Verhage 6 and 7 represent a high educational level (high level secondary education or university degree).23
Clinical Characteristics
Clinical information was retrieved from the electronic medical charts. Tumor location was classified as frontal (ie, frontal, frontal-temporal, frontal-parietal) vs nonfrontal (temporal, parietal, occipital, or a combi-nation of these areas), and according to lesion side (ie, right, left, bilateral) by means of a standard preoperative contrast-enhanced T1 weighted magnetic resonance image. Total preoperative tumor volume was segmented semiautomatically, followed by manual adjustments, with ITK-SNAP (www.itksnap.org)24or BrainLab (BrainLab, Munich,
Germany) software by trained researchers under supervision of the neurosurgeon. Karnofsky Performance Status (KPS) was reflected by the American Society of Anesthesiologist (ASA) score, ranging from ASA I (patient completely healthy) to ASA V (moribund patient) and considered dichotomous (ASA score I-II vs ASA score III-IV).25 Use
of psychotropic medication was defined as use of antiepileptic drugs, corticosteroid drugs, benzodiazepines, opioids, antipsychotics, stimu-lants, and/or antidepressants.
Psychological Characteristics
The Hospital Anxiety and Depression Scale (HADS; Dutch trans-lation) was used to assess self-reported symptoms of anxiety and depression.26,27 The HADS comprises 14 items: both subscales (ie,
anxiety and depression) include 7 items resulting in a score from 0 to 21 for both subscales. Higher scores represent more anxiety/depression symptoms, and in addition, the cut-off for clinical significance was set at 8 for each subscale.
Cognitive Performance
Cognitive functioning was assessed using the computerized battery Central Nervous System Vital Signs (CNS VS, Dutch translation).28,29
CNS VS includes 7 neuropsychological tests that reflect performance on the domains of Verbal Memory, Visual Memory, Processing Speed, Psychomotor Speed, Reaction Time, Complex Attention, and Cognitive Flexiblity.30Raw domain scores were converted into
sociodemographi-cally adjusted z scores, and scores at T3 were additionally corrected for effects of practice, since both sociodemographic and practice effects were demonstrated in a Dutch normative sample (age ranging from 20 to 80 yr, education ranging from 10 to 26 yr), assessed using CNS VS at baseline (n= 158), and 3-mo (n = 136) follow-up.28,29
It takes 30 to 40 min to complete CNS VS. Assessments were performed using the CNS VSX local software app, on a laptop computer running a 64-bit operating system. A well-trained test technician was present during each assessment.
Statistical Analyses
Patients’ Characteristics
Descriptive and comparative analyses (ie, between the patient sample that completed only T0 and the patients who completed both T0 and T3) of baseline sociodemographic, clinical, and psychological character-istics were performed using one-sample z tests and chi-square tests of independence.
Cognitive Performance and Changes on the Group Level
To examine potential differences in mean CNS VS performance on the 7 cognitive domains between GBM patients and the normative sample, 1-tailed one-sample z tests were performed for T0 and T3 (test values: mean (M) z= 0, standard deviation (SD) = 1). The mean
z score is comparable to the effect size (ES) Glass’s when calculated
as follows: Mpatients–Mcontrols/SDcontrols. Therefore, mean z scores≤ 0.50
were considered to represent small ES, between 0.51 and 0.79 medium ES, and≥0.80 large ES.31
We conducted 2-tailed paired samples t tests to assess changes in cognitive performance over time on the group level from T0 to T3. ESs were calculated and expressed as Cohen’s d as follows: Mdifference/SDdifference (d ≤ 0.50 = small, d between 0.51 and
0.80= medium, d ≥ 0.80 = large).31,32
Cognitive Performance and Changes on the Individual Patient Level
We counted the numbers and percentages of patients scoring impaired (ie, defined as a z score of≤1.50)33for all cognitive domains at both time
points.
Furthermore, we assessed changes in performance of individual patients over time by calculating reliable change indices (RCI) for each cognitive domain.29RCI values exceeding±1.645 represented changes,
whereby positive and negative values respectively represented improved and declined performance. Also, we counted the numbers of patients within every change category (improved, stable, and declined perfor-mance) for each domain.
Risk Factors for Postoperative Cognitive Impairment
Several potential preoperative risk factors were assessed: sociodemo-graphic (age, sex, educational level), clinical (tumor location in terms of hemisphere and frontal vs nonfrontal, ASA score, tumor volume), psychological (HADS anxiety and depression), and cognitive (T0 perfor-mance for each relevant cognitive domain) variables. A binominal logistic regression analysis was performed for each cognitive domain to examine the effect of these factors on the likelihood that patients scored impaired after surgery. Linearity of the continuous predictors with respect to the logit of the dependent variable was assessed via the Box-Tidwell procedure.34For every cognitive domain, the explained variance
Nagelkerke R square (R2),35 the percentage accuracy in classification
(PAC), sensitivity (ie, true positive rate), and specificity (ie, true negative rate) were presented. In addition, the receiver operating characteristic (ROC) curves including area under the curve values (AUC) were shown, representing the ability of a model to discriminate between patients with and without cognitive impairment.36Data on the risk factors were
presented as B coefficients and associated standard errors (SE). Odds ratios (ORs), representing the change in odds of scoring impaired after surgery for each increase in an unit of the predictor, are presented as well.37
Statistical analyses were performed using the Statistical Package for the Social Sciences (SPSS) version 24.0 (IBM Corporate Headquarters, Armonk, New York). To reduce the false discovery rate due to multiple statistical testing, P-values were set against a corrected alpha, using the Benjamini-Hochberg (BH) procedure.38,39
RESULTS
Patients’ Characteristics
Figure1shows the flow of GBM patients through the study. At T0, 208 patients were included. Thirty-five percent of the patients dropped out before T3 (mostly due to clinical deterio-ration or decease), resulting in 136 patients at T3. Table1shows
patients’ characteristics. There were no significant differences with regard to baseline sociodemographic, clinical, or psychological characteristics of the T0-only and T3 samples (P-values>.133). In order to include a preoperative predictor reflecting functional status, we looked into KPS scores of the patients.40 Yet, out of the 208 included patients, only 51 patients were found to have a preoperatively determined KPS score recorded in their electronic patient file. Of these patients, only 3 patients had a KPS score of
<80. Given the amount of missing data, and the limited spread
in scores, we chose not to include KPS score as a predictor in our analysis.
Cognitive Performance and Changes on the Group Level
GBM patients demonstrated significantly lower performance compared to normative controls on all cognitive domains at T0 and T3 (ps< BH-corrected alpha .05) (Table2). Worst perfor-mance (at both T0 and T3) was found for Reaction Time (ES −2.10 and −1.88, respectively) and Complex Attention (ES −2.26 and −1.60, respectively).
Paired-samples t tests revealed no significant changes in cognitive performance on the group level between T0 and T3, except for Complex Attention, for which performance improved significantly (t(111)= –2.85, P = .005) (see Table2). With Glass’
ranging from 0.04 to 0.27, ESs were small for all domains.
Cognitive Performance and Changes on the Individual Patient Level
Impaired performance occurred in 29% up to 55% of the patients across different domains at T0. At T3, 23% to 49% of the patients showed impaired scores across the 7 cognitive domains (Figure2).
Changes in performance of individual patients over time are shown in Figure3. Improvements occurred in 11% (both Verbal Memory and Visual Memory) up to 32% (Cognitive Flexibility) of the patients across the domains. A total of 6% (Processing Speed) to 26% (Reaction Time) of the patients showed declined performance between T0 and T3.
With up to 56% and 47% of the patients showing changed performance, improvements and declines were most frequent on Reaction Time and Cognitive Flexibility. Fewest changes were observed for the memory domains, with 81% and 80% of the patients demonstrating stable performance over time on Verbal and Visual Memory, respectively.
Risk Factors for Postoperative Cognitive Impairment
The logistic regression models reached statistical significance for all cognitive domains (ps< BH-corrected .05). The explained variance Nagelkerke R2 ranged from 35% (Psychomotor Speed) up to 52% (Cognitive Flexibility), and the models correctly classified 72% (Reaction Time) to 83% (Visual Memory) of the patients (Table 3). Figure 4 demonstrates ROC curves for the 7 cognitive domains: AUC values ranged from 0.81 (Verbal Memory) to 0.88 (Cognitive Flexibility). Sensitivity and
FIGURE 1.Flowchart of glioblastoma patients eligible for inclusion and follow-up.
TABLE 1. Baseline Characteristics of Glioblastoma Patients at T0 and T3
T0 (n= 208) T3 (n= 136) z/χ2 P
Sociodemographic characteristics
Age (yr): mean± SD (range) 58.5± 11.4 (18-81) 57.2± 11.9 (18-80) − 1.33 .184 Education (yr): mean± SD (range) 14.0± 3.6 (4-25) 14.2± 3.6 (4-25) 0.65 .517 Sex: female/male, n (%) 61 (29)/147 (71) 41 (30)/95 (70) 0.09 .768
Clinical characteristics at T0
Hemisphere: left/right/bilateral n(%) 73 (35)/134 (64)/1 (1) 50 (37)/85 (62)/1 (1) 0.26 .876 Frontal/non-frontal, n(%) 66 (32)/142 (68) 44 (32)/92 (68) 1.89 .170 Tumor volume (cm3): median (range) 36.7 (0.7-435.4) 35.3 (0.7-163.4) − 0.14 .889
ASA score: I/II/III n(%) 79 (38)/114 (55)/15 (7) 56 (41)/70 (52)/10 (7) 0.69 .707 Psychotropic medication: yes/no, n(%) 196 (94)/12 (6) 132 (97)/4 (3) 2.26 .133
Adjuvant therapy at T3
Concomitant RT+ ChT/RT/ChT/none – 114 (84)/17 (13)/2 (1)/3 (2)
Psychological characteristics at T0
Anxiety HADS: mean± SD (range)a 6.8± 4.3 (0-18) 6.8± 4.2 (0-16) 0.00 1.00
Above clinical cut-off n(%) 69 (39) 23 (17)
Depression HADS: mean± SD (range)a 4.8± 3.2 (0-13) 4.8± 3.0 (0-13) 0.00 1.00
Above clinical cut-off n(%) 41 (23) 25 (19)
ASA, American Society of Anesthesiologists;20ChT, chemotherapy; HADS, Hospital Anxiety and Depression Scale41; RT, radiotherapy.
aData missing T0= 30; T3 = 18.∗P< .05.
specificity ranged from 46% to 68% and 76% to 94% over domains, respectively. Only presurgical cognitive performances were significant risk factors for postoperative impairments on all cognitive domains, except for Visual Memory, where sex was also found to be a significant risk factor (ps< BH-corrected alpha .005) (Table 3). For all domains, higher presurgical cognitive performance was associated with a decreased likelihood of postop-erative cognitive impairment (ORs ranging from 0.36 to 0.70, representing 30% to 64% higher risk on cognitive impairment for each SD lower in the preoperative z score), and females had 6.50 times higher odds to exhibit postoperative Visual Memory impairments than males.
As more of half of the patients showed stable performance over time on 6 out of 7 domains that were assessed (ie, with the
exception of the Reaction Time domain, where 44% of the GBM patients showed stable performance over time), these numbers were not sufficient to carry out statistical prediction analysis on group membership (ie, improvement, stable, or declined cognitive performance) at 3-mo follow-up.
DISCUSSION
We evaluated pre- and postoperative cognitive functioning in a sample of GBM patients using the computerized neuropsy-chological battery CNS VS, and sought to present presurgical factors that enable the identification of patients who are at risk for postoperative cognitive deficits.
TABLE 2. Cognitive Performance on CNS VS of Glioblastoma Patients Pre- and Postoperatively
Cognitive domain Mean z score ± SD Na z test Pb Glassc
T0 preoperative NPA Verbal Memory − 0.90 ± 1.33 197 −12.58 <.001∗ −0.90 Visual Memory − 0.86 ± 1.29 204 −12.28 <.001∗ −0.86 Processing Speed − 1.31 ± 1.46 201 −18.63 <.001∗ −1.31 Psychomotor Speed − 1.62 ± 1.82 199 −22.82 <.001∗ −1.62 Reaction Time − 2.10 ± 2.71 193 −29.08 <.001∗ −2.10 Complex Attention − 2.26 ± 2.66 186 −30.57 <.001∗ −2.26 Cognitive Flexibility − 1.97 ± 2.26 187 −26.91 <.001∗ −1.97 T3 postoperative NPA Verbal Memory − 1.08 ± 1.58 129 −12.31 <.001∗ −1.08 Visual Memory − 0.64 ± 1.23 132 −7.39 <.001∗ −0.64 Processing Speed − 1.10 ± 1.34 133 −12.73 <.001∗ −1.10 Psychomotor Speed − 1.23 ± 1.65 134 −14.25 <.001∗ −1.23 Reaction Time − 1.88 ± 2.51 130 −21.41 <.001∗ −1.88 Complex Attention − 1.60 ± 2.61 122 −17.73 <.001∗ −1.60 Cognitive Flexibility − 1.56 ± 1.97 125 −17.49 <.001∗ −1.56
T0-T3 pairs Mean difference± SD N t test Pe Cohen’s dd
Verbal Memory − 0.23 ± 1.59 120 1.56 .121 −0.14 Visual Memory 0.14± 1.32 129 −1.20 .232 0.04 Processing Speed 0.16± 1.18 128 −1.51 .133 0.10 Psychomotor Speed 0.31± 1.73 129 −2.06 .042 0.18 Reaction Time 0.26± 2.46 120 −1.14 .255 0.11 Complex Attention 0.65± 2.42 112 −2.85 .005∗ 0.27 Cognitive Flexibility 0.33± 1.89 117 −1.87 .064 0.17
aThe number of patients differ over cognitive domains as a consequence of missing or invalid scores. b∗P value< BH-corrected alpha .05.
c,dGlass’s and Cohen’s d ES with ≤0.50 = small, 0.51-0.79 = medium, ≥= large.9,27
e∗P value< BH-corrected alpha .007.
Key Results
Consistent with the literature and our previous work, we found pre- and postoperative cognitive deficits to be very common in GBM patients, eg,6,11,12,14up to 55% and 49% of the patients showed impaired performance over the different domains prior to and after surgery, respectively. Besides very common, deficits also proved to be severe: ESs were found to be major in general (ES ranging from −0.86 to −2.26). Furthermore, our results indicate that improvements of postoperative performance are approximately as frequent as declines. Over time, only perfor-mance on Complex Attention improved significantly at the mean group level, yet postoperative group performance was still signif-icantly and greatly impaired (with an ES of −1.60). No signif-icant mean group changes over time were found for the other 6 cognitive domains that were assessed. Also, more than half of the patients showed stable performance over time on 6 out of the 7 domains that were assessed, suggesting that cognitive functioning does not necessarily significantly decline postsurgery. A recent meta-analysis on cognitive changes in glioma patients suggested a beneficial effect of surgery on cognitive functioning,10 yet several studies that were excluded reported no significant declines
after surgery at best, which is more in line with the current study.12,20,41,42 Furthermore, another recent meta-analysis on cognitive functioning specifically in patients with GBM reported a consistently high risk of cognitive dysfunction and further deterioration of cognitive functioning after surgical treatment (with 7 out of 11 studies reporting static deficits or deteriorated performance).43
Low preoperative cognitive performance significantly increased the odds for GBM patients with regard to showing postoper-ative deficits on all domains. In addition, female sex was a risk factor for postoperative deficits in Visual Memory. Significant overall male advantages have been described in the literature with regard to the storage of visual and spatial information—with an increase in the magnitude of sex differences with age of partici-pants.44Given the mean age of 59 yr in the current sample, the significant female disadvantage on Visual Memory in the current sample is in line with the literature. In order to further evaluate this finding, we examined potential differences between females and males in the current study: no significant differences were present with regard to age, education, hemispheric lesion side, frontal tumor involvement, ASA scores, tumor volume, anxiety,
FIGURE 2. Percentages of glioblastoma patients with impaired or nonimpaired performance on CNS VS cognitive domains at T0 and T3. The asterisk indicates a significantly larger proportion of patients with impaired performance as compared to in normative controls (ps< BH-corrected alpha .05).
and depression (data not shown), whereby it is unlikely that these variables explain the elevated risk on postoperative visual memory impairments in females. No other sociodemographic, clinical, or psychological factors were found to increase the risk of postoperative cognitive dysfunction. The literature on cognitive functioning in glioma patients with regard to the predictive value of lesion location and volume is not clear cut, and the finding that late cognitive outcomes do not vary by these variables is therefore not suprising.45-48 It has been suggested that gliomas can yield both local and global effects by infiltrating healthy brain tissue, reducing functional integrity of remote brain regions, and disturbing the cerebral network as a whole.4,9 Recently, even functional connectivity within the contralesional hemisphere has been found to play a role in determining the severity of cognitive impairments.47 Taken together, this may explain why tumor-related characteristics (ie, hemisphere, location, and volume) did not specifically predict outcomes for cognitive domains.
Interpretation
Various neuropsychological instruments have been used across studies to assess cognitive performance in GBM patients. Results
of the current and our previous study on cognitive impair-ments in GBM patients using a standardized computerized neuropsychological test battery (ie, CNS VS30) are very similar to the results of studies that examined patients using conven-tional paper-and-pencil tests.5,14,16,17 Therefore, computerized batteries could be useful tools both for research and clinical purposes, providing pre- and postoperative neuropsychological information on a fairly wide range of cognitive functions in a relatively short time. Yet, it should be mentioned that CNS VS is somewhat limited in terms of covering all relevant cognitive domains, as for example language and visuospatial abilities are not evaluated by its tests. Also, the CNS VS battery would benefit from a supplementary memory test (eg, addressing retrieval and learning efficiency), as its memory tests are constrained to recognition.
The current study shows that it is possible to identify patients at risk for postoperative cognitive impairments even before surgery. The risk of postoperative impairment over the different cognitive domains becomes 30% (Complex Attention) up to 64% (Processing Speed) higher for each unit lower (ie, in terms of SD) in the preoperative cognitive z score. Former studies have already provided evidence that cognitive perfor-mance is an important predictor of QoL and (progression
FIGURE 3. Percentages of glioblastoma patients with declined, stable, and improved performance at each cognitive domain across the 2 time intervals. The asterisk indicates a significantly larger proportion of changers as compared to in normative controls (ps< BH-corrected alpha .04).
free) survival.7-9Also, cognitive performance of glioma patients (assessed with neuropsychological tests) has been found to be generalizable to “real-world” functions and activities,49 resulting in for example slow responses to stimuli (related to Reaction Time) and struggles with adapting to new or changing events (related to Cognitive Flexibility), thereby complicating the ability of patients to perform everyday activities such as driving a car or preparing diner. We advocate the implementation of a preoper-ative and 3-mo postoperpreoper-ative NPA into the clinical care of GBM patients to gain insight into cognitive functioning and to guide in clinical decision-making. By doing so, a referral to, for example, cognitive rehabilitation can be provided timely with the aim of maintaining or even improving QoL and daily-life autonomy of patients. Additional research will be needed to examine methods for optimal rehabilitation of cognitive functioning (eg,
psychoe-ducation, strategy training, retraining, exercise interventions, pharmacological interventions) after surgery in GBM patients further, as promising results on cognitive rehabilitation in brain tumor patients have been demonstrated already.50
Limitations
The current study has some limitations that should be noted. We solely included patients who were considered acceptable candidates for surgery and capable of undergoing the NPA presurgery. Also, 35% of the patients dropped out before completion of the follow-up assessment, mainly (ie, in more than two-thirds of the patients) due to a poor clinical condition or decease of patients, which is inherent to the evolution of the disease. Consequently, results may be biased toward an overesti-mation of cognitive performance in GBM patients. Furthermore,
TA B L E 3 . Binomial L ogistic R egr e ssion P redic ting the Lik elihood of C o gnitiv e Impairmen t a t T 3 B ased on P reoper a tiv e S o ciodemogr aphic , C linical , P sy cholo gical , and C o gnitiv e V ariables V e rbal Memor y Vi su a l Memor y P roc essing Speed P sy chomotor Speed Reac tion Ti m e C o mple x At te n ti o n C o gnitiv e F le x ibility n 106 114 112 111 107 100 102 Nagelkerke R 2a 39.4 37.6 45.5 35.4 42.9 44.1 52.4 PA C b 81.1% 83.3% 80.4% 73.0% 72 .0% 78.0% 76 .5 % S e nsitivit y 54.4 4 6.2 6 5.9 56.8 6 7.9 6 4.1 6 4.1 Specificit y 93.2 94.3 88.7 83.6 75.9 86.9 84.1 V ariables B (SE) c OR d B (SE) c OR d B (SE) c OR d B (SE) c OR d B (SE) c OR d B (SE) c OR d B (SE) c OR d Ag e 0.02(.03) 1.02 − 0.02 (.03) 0.98 0.01 (.02) 1.01 0.00 (.02) 1.00 0.05 (.02) 1.05 0.07 (.03) 1.07 0.04 (.03) 1.04 Se x Fe m a le (v s m a le ) 0.02 (.56) 1.02 − 1.87 (.61) 6.50 0.53 (.55) 1.70 0.28 (.53) 1.32 − 0.07 (.56) 0.94 0.47 (.5 9) 1.60 1.12 (.63) 3.07 Education Lo w (v s o th e r) − 0.52 (.65) 0.60 0.49 (.67) 1.64 1.55 (.64) 4.70 − 0.17 (.5 9) 0.84 − 0.14 (.63) 0.87 − 0.70 (.65) 0.50 − 0.05 (.67) 0.95 H igh (v s o ther) − 0.97 (.63) 0.38 − 0.40 (.7 4) 0.70 1.16 (.64) 3.18 − 0.35 (.57) 0.7 1 − 0.03 (.58) 0.97 0.32 (.64) 1.38 − 0.07 (.6 9 ) 0.94 Hemispher e R ight (v s lef t) − 0.52 (.57) 0.60 − 0.26 (.65) 0.7 7 − 0.53 (.54) 0.5 9 0.05 (.52) 1.05 − 0.39 (.53) 0.68 0.31 (.57) 1.36 − 0.38 (.66) 0.6 9 Tu m o r v o lu m e 0.04 (.01) 1.00 0.01 (.01) 1.01 − 0.01 (.01) 0.99 − .01 (0.01) 1.00 − .01 (0.01) 1.00 0.02 (.01) 1.02 0.02 (.01) 1.02 Fr o n ta l Ye s (v s n o ) − 1.09 (.56) 2.99 − 0.68 (.60) 0.51 0.00 (.56) 1.00 0.43 (.51) 1.53 0.7 7 (.54) 2.16 1.12 (.57) 3.07 0.80 (.58) 2.23 ASA sc o re e > 3( v s1o r2 ) 1.38 (1.2 4) 3.95 1.57 (1.00) 4.80 1.7 6 (1.51) 5.83 0.82 (1.01) 2.27 1.2 4 (1.40) 3.47 1.2 2 (1.18) 3.38 0.91 (1.14) 2.49 HADS A f 0.06 (.08) 1.07 0.13 (.09) 1.14 0.08 (.08) 1.09 − 0.03 (.07) 0.97 0.12 (.08) 1.13 0.01 (.08) 1.01 − 0.01 (.08) 0.99 HADS D f − 0.1 9 (.12) 0.82 0.05 (.11) 1.05 − 0.02 (.10) 0.99 0.18 (.09) 1.1 9 0.04 (.10) 1.04 0.10 (.10) 1.11 0.12 (.11) 1.12 C og n itiv e sc or e T0 − 0.97 (.2 4 ) 0 .38 − 0.81 (.31) 0 .45 − 1.02 (.25) 0.3 6 − 0.7 4 (0.18) 0.48 − 0.54 (.1 4 ) 0.59 − 0.3 6 (.12) 0.70 − 0.7 1 (.16) 0.49 In bold: P < BH-corr e ct ed alpha .005. aNagelkerke R 2,l a rg e rR 2values indicating mor e varianc e explained b y the m odel ,t o a maximum o f 1. 29 bP e rc entage A cc ur ac y in C lassification (P A C ); the per centage o f the to tal sample that is co rr ec tly classified b y the model . cB coefficients and a ssociat ed standar d err ors . dOR fo r the pr edic tors ,O R smaller than 1.00 indicat e a d ecr eased odd o f impairment for an incr ease in one u nit o f the pr edic to r. 32 eASA, American S o ciet y o f A nesthesiolog ists . 20 fHADS, Hospital Anxiet y a nd Depr ession S cale . 21
FIGURE 4. ROC curves for sociodemographic, clinical, psychological, and cognitive risk factors predicting postoperative cognitive impairments in glioblastoma patients. AUC= area under the curve, representing the ability of a model to discriminate between patients with and without cognitive impairment, with values of <0.50, ≥0.51 to ≤0.70, ≥0.71 to ≤0.80, ≥0.81 to ≤0.90, and ≥0.91 suggesting no, poor, acceptable, excellent, or outstanding discrimination, respectively.30
patients were receiving adjuvant treatments at time of the 3-mo follow-up assessment. As this was the case in the far majority of the patients, we were unable to take effects of adjuvant radiotherapy and/or chemotherapy statistically into account in this study.
Yet, radiotherapy and chemotherapy represent an additional (ie, in addition to the tumor itself and surgical resection) risk for cognitive impairment, even with relatively well tolerated medica-tions such as temozolomide.51
Future Directions
It would be desirable to be able to identify patients who are at risk for cognitive impairment on the basis of information that can be collected with relatively little effort (eg, sociodemo-graphics, basic clinical data) at an early stage of disease. Yet, no preoperative sociodemographic (except for Visual Memory), clinical, or psychological factors were found to have significant predictive power with regard to cognitive outcomes of GBM patients at 3-mo follow-up. Future studies should aim to assess the potential added predictive value of factors that are known after surgery (eg, additional radiotherapy and/or chemotherapy, isocitrate dehydrogenase status, and disease progression) on later cognitive outcomes. By doing so, it can be determined whether cognitive outcomes of GBM patients can be predicted better if postoperative factors are added to prediction models, and moreover, risk factors for cognitive impairment that are only known postoperatively can be identified.
CONCLUSION
The current study shows that it is possible to identify patients at risk for postoperative cognitive impairments even before surgery. This information can help to inform patients and clinicians at an early stage, and emphasizes the importance of recognizing, assessing, and actively dealing with cognitive functioning in the clinical management of GBM patients.
Disclosures
This study is funded by ZonMw, a Dutch national organization for Health Research and Development (project number 824003007), and CZ Group, a Dutch non-profit health insurer’s foundation (project number 201500028). The authors have no personal, financial, or institutional interest in any of the drugs, materials, or devices described in this article.
REFERENCES
1. Weller M, Wick W, Aldape K, et al. Glioma. Nat Rev Dis Primers. 2015;16(1):15017.
2. Louis DN, Perry A, Reifenberger G, et al. The 2016 world health organization classification of tumors of the central nervous system: a summary. Acta Neuropathol. 2016;131(6):803-820.
3. Ostrom QT, Gittleman H, Truitt G, Boscia A, Kruchko C, Barnholtz-Sloan JS. CBTRUS statistical report: primary brain and other central nervous system tumors diagnosed in the united states in 2011-2015. Neuro Oncol. 2018;20(Suppl_4):1-86.
4. Lapointe S, Perry A, Butowski NA. Primary brain tumours in adults. The Lancet. 2018;392(10145):432-446.
5. Tykocki T, Eltayeb M. Ten-year survival in glioblastoma. A systematic review.
J Clin Neurosci. 2018;54:7-13.
6. Van Loenen I, Rijnen SJM, Bruijn J, Rutten GJM, Gehring K, Sitskoorn MM. Group changes in cognitive performance after surgery mask changes in individual patients with glioblastoma. World Neurosurg. 2018;117:e172-e179.
7. Drewes C, Sagberg LM, Jakola AS, Solheim O. Perioperative and postoper-ative quality of life in patients with glioma - a longitudinal cohort study. World
Neurosurg. 2018;117:e465-e474.
8. Johnson DR, Sawyer AM, Meyers A, O’Neill BP, Wefel JS. Early measures of cognitive function predict survival in patients with newly diagnosed glioblastoma.
Neuro Oncol. 2012;14(6):808-816.
9. Peters KB, Woodring S, Healhy P, et al. Baseline cognitive function to predict survival in patients with glioblastoma. J Clin Oncol. 2017;34(15_Suppl):10125.
10. Ng JCH, See AAQ, Ang TY, Ang BT, King NKK. Effects of surgery on neurocognitive function in patients with glioma: a meta-analysis of immediate post-operative and long-term follow-up neurocognitive outcomes. J Neurooncol. 2018;141(1):167-182.
11. Raysi Dehcordi S, Mariono M, Mazza M, Galzio RJ. Cognitive deficits in patients with low and high grade gliomas. J Neurosurg Sci. 2013;57(4):259-266. 12. Habets EJ, Kloet A, Walchenbach R, Vecht CJ, Klein M, Taphoorn MJ. Tumour
and surgery effects on cognitive functioning in high-grade glioma patients. Acta
Neurochir (Wien). 2014;156(8):1451-1459.
13. Habets EJ, Taphoorn MJ, Nederend S, et al. Health-related quality of life and cognitive functioning in long-term anaplastic oligodendroglioma and oligoastro-cytoma survivors. J Neurooncol. 2014;116(1):161-168.
14. Van Kessel E, Baumfalk AE, van Zandvoort MJ, Robe PA, Snijders TJ. Tumor-related neurocognitive dysfunction in patiens with diffuse glioma: a systematic review of neurocognitive functioning prior to anti-tumor treatment. J Neurooncol. 2017;134(1):9-18.
15. Ali FS, Hussain MR, Gutiérrez C, et al. Cognitive disability in adult patients with brain tumors. Cancer Treat Rev. 2018;65:33-40.
16. Hendrix P, Hans E, Griessenauer CJ, Simgen A, Oertel J, Karbach J. Neurocog-nitive status in patients with newly-diagnosed brain tumors in good neurological condition: the impact of tumor type, volume, and location. Clin Neurol Neurosurg. 2017;156:55-62.
17. Dallabona M, Sarubbo S, Merler S, et al. Impact of mass effect, tumor location, age, and surgery on the cognitive outcome of patients with high-grade gliomas: a longitudinal study. Neurooncol Pract. 2017;4(4):229-240.
18. Gempt J, Lange N, Bette S, et al. Factors influencing neurocognitive function in patients with neuroepithelial tumors. Sci Rep. 2017;7:17764.
19. Flechl B, Sax C, Ackerl M, et al. The course of quality of life and neurocognition in newly diagnosed patients with glioblastoma. Radiother Oncol. 2017;125(2):228-233.
20. Hendriks EJ, Habets EJJ, Taphoorn MJB, Douw L, et al. Linking late cognitive outcome with glioma surgery location using resection cavity maps. Hum Brain
Mapp. 2018;39(5):2064-2074.
21. Solanki C, Sadana D, Arimappamagan A, et al. Impairment of quality of life and cognitive functions in long-term survivors of glioblastoma. J Neurosci Rural Pract. 2017;8(2):228-235.
22. Bergo E, Lombardi G, Guglieri I, et al. Neurocognitive functions and health-related quality of life in glioblastoma patients: a concise review of the literature. Eur J
Cancer Care. 2015;28(1):e12410.
23. Verhage F. Intelligentie en leeftijd onderzoek bij Nederlanders van twaalf tot
zeven-zeventig jaar. [Intelligence and age: Research study in Dutch individuals aged twelve to seventy-seven]. Assen, The Netherlands: Van Gorcum/Prakke & Prakke; 1964.
24. Yushkevich PA, Gerig G. ITK-SNAP: an interactive medical image segmentation tool to meet the need for expert-guided segmentation of complex medical images.
IEEE Pulse. 2017;8(4):54-57.
25. Dripps RD. New classification of physical status. Anesthesiology. 2008;24(111). 26. Zigmond, AS, Snaith, RP. The Hospital Anxiety and Depression Scale. Acta Psychiat
Scand. 1983;67(6):361-370.
27. Spinhoven P, Ormel J, Sloekers PP, Kempen GI, Speckens AE, van Hemert AM. A validation study of the Hospital Anxiety and Depression Scale (HADS) in different groups of Dutch subjects. Psychol Med. 1997;27(2):363-370.
28. Rijnen SJM, Meskal I, Emons WHM, et al. Evaluation of normative data of a widely used computerized neuropsychological battery: applicability and effects of sociodemographic variables in a Dutch sample. Assessment. 2017;27(2):373-383.
29. Rijnen SJM, Van der Linden SD, Emons WHM, Sitskoorn MM, Gehring K. Test-retest reliability and practice effects of a computerized neuropsychological battery: a solution-oriented approach. Psychol Assess. 2018;30(12):1652-1662.
30. Gualtieri CT, Johnson LG. Reliability and validity of a computerized neurocog-nitive test battery, CNS vital signs. Arch Clin Neuropsychol. 2006;21(7):623-643. 31. Glass GV, McGaw B, Smith ML. Meta-analysis in Social Research. Beverly Hills:
Sage Publications, 1981.
32. Cohen J. In: Statistical Power Analysis for the Behavioral Sciences. 2nd edn. Hillsdale, NJ: Lawrence Erlbaum Associates; 1988.
33. Lezak MD, Howieson DB, Bigler ED, Trandel D. Neuropsychological Assessment, 5 edn. New York, NY: Oxford University Press, 2012.
34. Box GEP, Tidwell PW. Transformation of the independent variables.
Technometrics. 1962;4(4):531-550.
35. Nagelkerke NJD. A note on the general definition on the coefficient of determi-nation. Biometrika. 1991;78(3):691-692.
36. Hosmer DW Jr, Lemeshow S, Sturdivant RX. Applied Logistic Regression, 3rd edn. Hoboken, NJ: Wiley, 2013.
37. Grimes DA, Schulz KF. Making sense of odds and odds ratios. Obstet Gynecol. 2008;111(2 pt 1):423-426.
38. Benjamini Y, Hochberg Y. Controlling the false discovery rate: a practical and powerful approach to multiple testing. J R Stat Soc Ser B. 1995;57(1):189-300.
39. Rijnen SJM, Meskal I, Bakker M, et al. Cognitive outcomes in meningioma patients undergoing surgery: individual changes over time and predictors of late cognitive functioning. Neuro Oncol. 2019;21(7):911-922.
40. Karnofsky DA, Burchenal JH. The clinical evaluation of chemotherapeutic agents in cancer. In: Macleod CM, ed. Evaluation of Chemotherapeutic Agents. New York: Columbia University Press, 1949: 191-205.
41. Noll KR, Weinberg JS, Ziu M, Benveniste RJ, Suki D, Wefel JS. Neurocognitive changes associated with surgical resection of left and right temporal lobe glioma.
Neurosurgery. 2015;77(5):777-785.
42. Rijnen SJM, Sitskoorn MM, Gehring K. Comment on: effects of surgery on neurocognitive function in patients with glioma: a meta-analysis of immediate post-operative and long-term follow-up neurocognitive outcomes. J Neurooncol. 2019;143(1):175-176.
43. Sinha R, Stephenson JM, Price SJ. A systematic review of cognitive function in patients with glioblastoma undergoing surgery. Neurooncol Pract. 2020;7(2):131-142.
44. Voyer D, Voyer SD, Saint-Aubin J. Sex differences in visual-spatial working memory: a meta-analysis. Psychon Bull Rev. 2017;24(2):307-334.
45. Habets EJJ, Hendriks EJ, Taphoorn MJB, et al. Association between tumor location and neurocognitive functioning using tumor localization maps.
J Neurooncol. 2019;144(3):573-582.
46. Hendriks EJ, Habets EJJ, Taphoorn MJB, et al. Linking late cognitive outcome with glioma surgery location using resection cavity maps. Hum Brain Mapp. 2018;39(5):2064-2074.
47. Dallabona M, Sarubbo S, Merler S, et al. Impact of mass effect, tumor location, age, and surgery on the cognitive outcome of patients with high-grade gliomas: a longitudinal study. Neurooncol Pract. 2017;4(4):229-240.
48. De Baene W, Rutten GM, Sitskoorn MM. Cognitive functioning in glioma patients is related to functional connectivity measures of the non-tumoural hemisphere. Eur J Neurosci. 2019;50(12):3291-3933.
49. Noll KR, Bradshaw ME, Weinberg JS, Wefel JS. Neurocognitive functioning is associated with functional independence in newly diagnosed patients with temporal lobe glioma. Neurooncol Pract. 2018;5(3):184-193.
50. Van Lonkhuizen PJC, Klaver KM, Wefel JS, Sitskoorn MM, Schagen SB, Gehring K. Interventions for cognitive problems in adults with brain cander: a narrative review. Eur J Cancer Care. 2019;28(3):e13088.
51. Noll KR, Bradshaw ME, Parsons MW, Dawson EL, Rexer J, Wefel JS. Monitoring of neurocognitive function in the care of patients with brain tumors. Curr Treat
Options Neurol. 2019;21(7):33.
